Limits to the chemical background and the mobility-selected current transmitted in a Differential mobility analyzer (DMA)

Mario Amo,1 Juan Fernández de la Mora2

1SEADM S. L., Boecillo, Spain; 2Yale University, Mech. Eng. Dept., New Haven, CT 06520-8286, USA

62th ASMS Conference on Mass Spectrometry and Allied Topics – Baltimore MD, MN – June 5-19, 2014

QUALITY OF MOBILITY PEAKS

CONCLUSIONSDMAs can be operated in clean conditions avoiding solvation

tails, with a fast decay of the peaks typically by 4 orders of

magnitude or more.

The DMA naturally reduces space charge problems by selecting

only a narrow range of the ions passed into the MS. It is also

capable of transmitting nA currents of mobility-selected ions.

ACKNOWLEDGEMENTSWe are grateful to Dr. Anatoly Verenchikov for many insightful suggestions on the ideal

coupling of a DMA to a mass spectrometer.

References:

[1]: J. Rus; D. Moro; J.A. Sillero; J. Royuela, A. Casado, J. Fernández de la Mora, IMS-MS studies based on coupling a Differential Mobility Analyzer (DMA) to commercial API-MS systems, Int. J. Mass Spectrom, 298, 30-40 (2010)

[2] Rosell, J., I. G. Loscertales D. Bingham and J. Fernández de la Mora "Sizing nanoparticles and ions with a short differential mobility analyzer", J. Aerosol Science, 27, 695-719, 1996.

[3] J. D. Cole. On a quasi-linear parabolic equation occurring in aerodynamics, Quart. Applied Math. 9 (3): 225-236 (1951).

[4] J. Fernandez de la Mora, The spreading of a charged cloud, Burgers' equation, and nonlinear simple waves in an ideal gas, pp. 250-257 in Simplicity, Rigor and Relevance in Fluid Mechanics, F.J. Higuera, J. Jiménez and J.M. Vega (Eds.), Barcelona, Spain 2004

Psty 9,2k 326 μM in 40%DMAF in NMP:DMF 1:1

+1

+4

+1

+3

+4

+5

-1

-2

-3

-4

-5-6

-7

-8

-9

+1

+2

+3

+4+5

OVERVIEW

The purity of the sheath gas was studied in a parallel-plate DMA (SEADM, model P5-e) interfaced to Sciex’s QTRAP-5500. The coupling permits fast removal of the DMA without breaking the vacuum. The ion source was an electric discharge in N2. (Fig 2).

CLEANNESS OF DMA CIRCUIT AND DYNAMIC RANGE

Differential mobility analyzers (DMAs) installed in the ion source region of API-MS instrumentscan convert a preexisting MS into an IMS-MS instrument [1]. Here we investigate:

1) DMA cleanness: DMAs require a high flow rate of sheath gas (~1000 lit/min) moved by amechanical blower, where high gas purity appears as harder to achieve than inconventional drift-tube mobility cells. Here we show that high gas purity is achievable withexisting commercial blowers. This cleanness leads to sharp mobility peaks withoutsolvation tails, enabling high quality ion separation

2) The claimed high ion transmission efficiency of DMAs has not been investigated insituations with high space charge, as when a nanoES source is brought very near the inletslit of the DMA. Here we demonstrate the ability to transmit ~ 1 nA of mobility-selectedions to the MS.

Fig. B1: Mobility spectrum for aspray of 100 mM Ethyl3N-Formate (TEAF) in methanol,demonstrating that most of thecharge is in the form of a singleion. Electrometer amplification:Green, 1nA/V; Blue: 0.1 nA/V.Needle ~several cm away fromthe inlet slit of the DMA.Experiment in CO2 (position ofvarious ions may differ fromthose later seen in air).

Fig 2B: Mobility peaks under conditions of increasing space charge. At low

currents (left) the peak is Gaussian, with a diffusion-controlled width independent

of ion current. At increasing currents (~0.25 nA) the peak height ceases to

increase, while its width increases drastically due to space charge broadening of

the ion beam. The low mobility tail on the right is due to incompletely desolvated

ions. These are promoted at higher ion fluxes because the current is increased by

bringing the ES tip closer to the inlet slit of the DMA.

Theory on space charge plus diffusion broadeningThe theory exploits the smallness of diffusion effects to convert the

problem into a 2D problem in space and time, similarly as in [2]. This

leads to Burgers equation, which can be solved analytically via the Cole

Hopf transformation [3], with further details given in [4]. We assume that

the initial ion beam is infinitely thin.

Fig. 2: SEADM’s clean corona

source for DMA

EXPERIMENTAL

Ion transmission studies used a DMA similar to that in Fig. 1, with an ES

chamber producing exclusively Eth3N+ ions. To control the ion current ingested

and transmitted by the DMA, the sharpened silica ES needle could be brought

arbitrarily close to the inlet slit of the DMA. The current of Eth3N+ conveyed by

the sampled flow rate of 0.6 lit/min was measured in a faraday cup

electrometer. The shape of the mobility peak was obtained by scanning the

voltage difference between both DMA plates.

Fig. 3: Left: DMA-MS spectrum. Right, mass spectrum including all mobilities.

With the clean corona source, the dominant peaks are associated to electron

ionizationa of TEFLON vapors (from teflon O-rings; m/z peak series spaced by the 50

Da characteristic of CF2). Given the exceptionally low outgassing rates of teflon, we

conclude that a high level of cleanness has been reached in the DMA circuit. The

blower is therefore not a source of vapor impurities!

a Electrons can be readily eliminated when desired

Solvation problemFig. 4: DMA-MS spectra of nitroglycerin (NG) vapors released by a NG pill: Blue: Clean system with negative corona ionization and no vapor impurities: Sharp tail decay (289 = 227NG +62NO3)Red: Traces of Methanol vapors admited in the chamber lead to clustering tails (262 = 227NG + 35Cl).

Left tails are due to electron attach-ment taking place inside the DMA

DMA voltage (Volt)

Peak quality level & dynamic range in clean system

Maximum mobility-selected current transmitted by a DMAThe DMA receives a mixture of ions at its inlet slit, and filters out all except those

whose electrical mobilities are within a narrow range centered at a controllable

value.

THE QUESTION is how high a current of mobility-selected ions can be transmitted

given the high space charge conditions typical of a nanospray plume. The DMA

immediately separates out the analytically interesting ions (of intermediate

mobilities) from the dominant space charge sources, including high mobility buffer

ions (such as ammonium+ or acetate-), as well low mobility clusters or incompletely

dried drops. This separation suffices in most situations to remove space charge

effects from the analytically relevant ions. However, a space charge limit exists for

the dominant ions in the plume, and it is this upper limit to the current which we

propose to study here. This space charge leads to lateral broadening of the ion

beam within the DMA, with important negative consequences on beam dilution and

loss of mobility resolution

m/z

(Da)

MATERIALS USED

Pump: rotor an stator, Al. shaft seal: plastic.DMA Circuit: SS 304 + Teflon O-ringsDMA. SS 316 + Peek + Teflon O-ringsDMA-MS Interface: SS+ Teflon O-ringsIonization Source: S S316 + Macor + Al gaskets, W (needle), SS 316 (fittings).Inlet line: fused sillica-line stainless steel tubing.Outlet line: stainless Steel 316 tubing.Rotameters: Glass, SS, Teflon, and Viton o-rings.Nitogen purity: >99,999

RESULTS ON DMA

CIRCUIT CLEANNESS

Figure 7.1: Mobility scan for m/z61. X-axys DMA Voltage (V), Y-axis intensity (cps).

Figure 7.2: Mobility scan for m/z75. X-axys DMA Voltage (V), Y-axis intensty (cps).

Figure 7.3: Mobility scan for m/z 267. X-axys DMA Voltage (V), Y-axis intensty (cps).

DMA Voltage (kV)

Fig. 1: Sketch of

experimental system with

clean DMA circuit

Fig. 3B: Herethe low mobility tail associated to incomplete desolvation is removed by

heating up the gas to 50 oC. The resulting peaks are now symmetric, as predicted by

theory including diffusion and space charge effects.

Figure 4B: Comparison of predicted(black points) versus observed peak

shapes, with ion current decreasing from top to bottom and left to right.

The ion diffusivity is taken to be that providing a best fit for the data at

low currents rather than that inferred from the measured electrical

mobility

distance in reality is limited by the need for the ES drops to evaporate and release the ions, and

for ion solvation to be minimized by a good drying process.

It is instructive to compare the predicted peak shapes (Figure A1) with those observed (Figure

A3).

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

"

___________________________________________________________________

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

Figure A4: Comparison of experimental and theoretical peak shapes for the data of Figure 13.7.

Calculated lines (red) correspond from wider to narrower to b = u¥(t/4D)1/2

= 4.4, 3.9, 3.05, 2.75,

2.37, 1.69, 0.95, 0.65, 0, s uperposition of all data.

We have found the coefficient relating linearly the theoretical variable a and the experimental

voltage variable V-Vo by fitting the peak next to lowest on Figure A3 to the Gaussian Exp[-a2].

The other peaks are fitted to the best b value, with the level of agreement shown in Figure A4.

The experimental tails seen on the right of the peaks are noteworthy. The fact that they are much

weaker than the tails observed under conditions of less effective drying suggests strongly that

they are due to solvation, which evidently increases as the needle is brought closer to the DMA

inlet in order to increase space charge effects. Except for this right tail anomaly, the fit is

generally excellent. Nonetheless, for the three cases with highest space charge, the experimental

data are slightly more curved at the top than the theoretical curve. A perfect fit cannot in any

case be expected due to the inexact nature of the comparison. First, the parameter a is not

computed from first principles, but is rather inferred by fitting one of the curves, assumed to

distance in reality is limited by the need for the ES drops to evaporate and release the ions, and

for ion solvation to be minimized by a good drying process.

It is instructive to compare the predicted peak shapes (Figure A1) with those observed (Figure

A3).

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

"

___________________________________________________________________

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

Figure A4: Comparison of experimental and theoretical peak shapes for the data of Figure 13.7.

Calculated lines (red) correspond from wider to narrower to b = u¥(t/4D)1/2

= 4.4, 3.9, 3.05, 2.75,

2.37, 1.69, 0.95, 0.65, 0, s uperposition of all data.

We have found the coefficient relating linearly the theoretical variable a and the experimental

voltage variable V-Vo by fitting the peak next to lowest on Figure A3 to the Gaussian Exp[-a2].

The other peaks are fitted to the best b value, with the level of agreement shown in Figure A4.

The experimental tails seen on the right of the peaks are noteworthy. The fact that they are much

weaker than the tails observed under conditions of less effective drying suggests strongly that

they are due to solvation, which evidently increases as the needle is brought closer to the DMA

inlet in order to increase space charge effects. Except for this right tail anomaly, the fit is

generally excellent. Nonetheless, for the three cases with highest space charge, the experimental

data are slightly more curved at the top than the theoretical curve. A perfect fit cannot in any

case be expected due to the inexact nature of the comparison. First, the parameter a is not

computed from first principles, but is rather inferred by fitting one of the curves, assumed to

distance in reality is limited by the need for the ES drops to evaporate and release the ions, and

for ion solvation to be minimized by a good drying process.

It is instructive to compare the predicted peak shapes (Figure A1) with those observed (Figure

A3).

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

"

___________________________________________________________________

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

Figure A4: Comparison of experimental and theoretical peak shapes for the data of Figure 13.7.

Calculated lines (red) correspond from wider to narrower to b = u¥(t/4D)1/2

= 4.4, 3.9, 3.05, 2.75,

2.37, 1.69, 0.95, 0.65, 0, s uperposition of all data.

We have found the coefficient relating linearly the theoretical variable a and the experimental

voltage variable V-Vo by fitting the peak next to lowest on Figure A3 to the Gaussian Exp[-a2].

The other peaks are fitted to the best b value, with the level of agreement shown in Figure A4.

The experimental tails seen on the right of the peaks are noteworthy. The fact that they are much

weaker than the tails observed under conditions of less effective drying suggests strongly that

they are due to solvation, which evidently increases as the needle is brought closer to the DMA

inlet in order to increase space charge effects. Except for this right tail anomaly, the fit is

generally excellent. Nonetheless, for the three cases with highest space charge, the experimental

data are slightly more curved at the top than the theoretical curve. A perfect fit cannot in any

case be expected due to the inexact nature of the comparison. First, the parameter a is not

computed from first principles, but is rather inferred by fitting one of the curves, assumed to

distance in reality is limited by the need for the ES drops to evaporate and release the ions, and

for ion solvation to be minimized by a good drying process.

It is instructive to compare the predicted peak shapes (Figure A1) with those observed (Figure

A3).

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

"

___________________________________________________________________

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 5 5 10

0.2

0.4

0.6

0.8

1.0

! 10 ! 5 5 10

0.2

0.4

0.6

0.8

1.0

Figure A4: Comparison of experimental and theoretical peak shapes for the data of Figure 13.7.

Calculated lines (red) correspond from wider to narrower to b = u¥(t/4D)1/2

= 4.4, 3.9, 3.05, 2.75,

2.37, 1.69, 0.95, 0.65, 0, s uperposition of all data.

We have found the coefficient relating linearly the theoretical variable a and the experimental

voltage variable V-Vo by fitting the peak next to lowest on Figure A3 to the Gaussian Exp[-a2].

The other peaks are fitted to the best b value, with the level of agreement shown in Figure A4.

The experimental tails seen on the right of the peaks are noteworthy. The fact that they are much

weaker than the tails observed under conditions of less effective drying suggests strongly that

they are due to solvation, which evidently increases as the needle is brought closer to the DMA

inlet in order to increase space charge effects. Except for this right tail anomaly, the fit is

generally excellent. Nonetheless, for the three cases with highest space charge, the experimental

data are slightly more curved at the top than the theoretical curve. A perfect fit cannot in any

case be expected due to the inexact nature of the comparison. First, the parameter a is not

computed from first principles, but is rather inferred by fitting one of the curves, assumed to

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.00 1.10 1.20 1.30 1.40 1.50 1.60 1.70 1.80

2

4

6

7

y = 0.6253xR² = 0.9712

0

0.5

1

1.5

0 0.5 1 1.5 2 2.5

Sample flow rate (lit/min)

Imax (nA)

Figure 5 B: Effect of increasingsample outlet flow rate qo: 0.5,0.75, 1, and 1.5 lit/min of sampleflow. The approximate nonlinearityin the dependence of the peaksignal in the flow rate suggeststhat the loss of ions between DMAoutlet and electrometer detectormust be modest.

The effectiveness of the DMA in simplifying complex mass spectra

depends not just on a narrow mobility peak, but on the complete

absence of small contaminating tails, particularly those due to solvation.

We therefore define the dynamic

range as illustrated in the figure on the

right, and provide various illustrations

of mobility peaks demonstrating sharp

decay of the peaks without solvation

tails, with dynamic ranges in excess of

104

Fig. 6: Mobility scan for m/z48. X-axis DMA Voltage (V), Y-axis intensity

EFFECT OF SAMPLE FLOW RATE ON MAXIMAL ION CURRENT

Fig. 5: Dynamic Range definition